Building Self-Accelerating AI with Mirendil

**Behnam Neyshabur** (0:00)
With an AI technology that is helping with AI research, what is the role of everyone else in the company? This is a very disruptive technology. With disruptive technologies, you need to rethink a lot of pieces to kind of enable them to actually grow and flourish.
And some of these pieces are how you build a company around it. If the business model of the company is, I train a big model and charge people for using it, how is this company incentivized to share this technology with everyone else?

**Harsh Mehta** (0:30)
We've been at the Frontier Labs for a long time and we know how much work it is to do something and we've been able to do it in like maybe 10 times less people and less resources. The jump from Sonar 3.5, 4 Opus and 4.5 has been materially better in terms of like where does it get stuck? Where does it need oversight? Even the tiniest reduction in oversight can lead to like large amounts of token spent and just effective outcomes being generated.

**Matt Bornstein** (0:57)
What you guys are doing is almost the next level of cheat code. When the Internet gets faster, you don't like recursively get faster Internet. Where do you think it all ends up?

**Harsh Mehta** (1:11)
It's a wide spectrum of outcomes.

**SPEAKER_4** (1:15)
One of the oldest ideas in AI is also one of the most ambitious. The idea that intelligent systems could help improve themselves.
For decades, that possibility remained largely theoretical. But as AI systems become more capable at coding, research, reasoning and engineering, the question is becoming increasingly practical. What would it mean to build AI systems that contribute to their own development? And if that becomes possible, where should that capability be directed? Matt Bornstein speaks with Mirendil co-founders, Behnam Neyshabur and Harsh Mehta, about self-accelerating AI, scientific discovery, and why they believe the most important applications of AI may be in advancing science itself.

**Matt Bornstein** (2:01)
Behnam, Harsh, welcome to the a16z podcast. It's awesome to have you here. Behnam and Harsh were most recently research scientists at Anthropic, worked together before at Google at BlueShift Labs, which Behnam, you were a founding member of. Thrilled to have you here. Tell us a little bit about Mirendil, what you're aiming to accomplish, and tell us where the company came from.

**Behnam Neyshabur** (2:18)
A lot of reasons behind building the company comes back from when scaling law was happening at OpenAI, and back then I was at BlueShift team, and that was the moment when I realized we are on the verge of a revolution, and then everything is about to change. And for me, the most important thing was, what are the technologies that would accelerate all areas of science, and what are the main bottlenecks for making that happen. And since then, I've been trying to remain close to that path and remain on the shortest path, and known Harsh since then, and trying to be on that path. And more recently, as we've been in Anthropic, we've talked a lot about what does it mean to kind of remain on this shortest path. And we felt like the main labs are starting to diverge a little bit from what does it take to accelerate all areas of science that led to starting Mirendil. We think the self-accelerating AI is a technology that is disruptive and at the same time, it's important for accelerating science and technology. And we want to focus on building that and making it available.

**Matt Bornstein** (3:23)
So science is notoriously hard, right? I mean, science is science. It's an experimental discipline where you have to run experiments in the real world to understand the laws of physics or biology or so forth. Can you guys describe a little bit, what do you think is the shortest path for AI to accelerate science and why?

**Harsh Mehta** (3:41)
So from the last five years, when the models started getting better at some of the primitives of connecting science, like high school math and college math, and then a little bit of coding, and then now the coding models are really good, competitive use, to start a good really.

**Matt Bornstein** (3:56)
Is this all in scope when you say science, by the way? I just want to make sure, are we talking about strictly physical natural sciences or is it a broader set of things?

**Harsh Mehta** (4:04)
So ultimately, what we want to build is AI systems, which in a very broad sense can conduct AI research and engineering itself. If it has this capability in a very broad sense, it has the right primitives and the capabilities which are needed to conduct any science which is in the realm of digital world or any science which is in the physical world, but then has a digital component as well.